Slashdot Mirror
Budweiser, the World's Largest Beer Maker, is Using Low-Cost Sensors and Machine Learning To Keep Beverages Flowing (wsj.com)
The world's largest beer maker is using low-cost sensors and machine learning to predict when motors at a Fort Collins, Colo. brewery might malfunction. From a report: The Anheuser-Busch InBev SA plant was the first among the company's 350 beverage-making facilities to test whether wireless sensors that can detect ultrasonic sounds -- beyond the grasp of the human ear -- can be analyzed to predict when machines need maintenance. "You can start hearing days in advance that something will go wrong, and you'll know within hours when it'll fail. It's really, for us, very practical," said Tassilo Festetics, vice president of global solutions for the company.
The project began about six months ago when Mr. Festetics's team installed 20 wireless sensors across three packaging lines motors to measure vibrations. The sounds picked up are transmitted in real time and then compared to a normal, functioning engine's sounds, which serve as a baseline and allow the program to identify anomalies. A key advantage is that the sensors are non-invasive and don't need to be placed inside a machine. Sensors have been used for predictive maintenance in the past, but they were unable to transmit information in real time. Advances in processing data at the edge of the network, referred to as edge computing, enables companies to collect and analyze real-time sensor data from machines.
The project began about six months ago when Mr. Festetics's team installed 20 wireless sensors across three packaging lines motors to measure vibrations. The sounds picked up are transmitted in real time and then compared to a normal, functioning engine's sounds, which serve as a baseline and allow the program to identify anomalies. A key advantage is that the sensors are non-invasive and don't need to be placed inside a machine. Sensors have been used for predictive maintenance in the past, but they were unable to transmit information in real time. Advances in processing data at the edge of the network, referred to as edge computing, enables companies to collect and analyze real-time sensor data from machines.
I never liked beer until I was in my 30s when the Craft Beer started to be commonly available in the States. Then I realized what I was tasting before was just crap.
Before when offered a beer I would just politely refuse.
If something is so important that you feel the need to post it on the internet... It probably isn't that important.
I am not familiar with the term edge computing. After a brief 2 Google searches, I'm skeptical this is isn't just a buzzword. Can anyone elaborate?
sensors that tell you when brewed Budweiser will taste terrible.
Wait a minute...
It's nice to control processes to turn out a consistent product. Well, unless it's Budweiser or Miller beer.
I often enjoy a local brewery that's been in business for over 140 years, (with the exception of prohibition years) and their prices are better, and their beer (not to mention Octoberfest and bock beers) are so much better than the "Major" Breweries like BW and Miller. Especially on tap, much less in cans and bottles.
The AE-35 dispensing unit is going to fail, Dave. You should go down to the factory floor and replace it before failure which would cause a large beer spill. I'll shut down the line for you Dave. It's perfectly safe.
Companies never post these kind of PR releases and journalists never do these stories without some payola.
I'm not a WSJ subscriber so screw getting past the paywall but rest assured that the technology provider was the instigator of this article so that they could create some FOMO energy in the marketplace. FOMO is "Fear of Missing Out". When other beverage companies read this, the executives will go "Oh Noes! We can't be left behind!" and hastily put a project together and call this provider. It's been happening forever. Nothing to see here, move along.
Now that Cloud Computing is the "new normal" we now have a new term for what used to be "computers in your fucking building". We call that "edge" computing. All that was old is new again!
Sorry weak beer faggot TRUMP TRAITORS!
You can start hearing days in advance that something will go wrong
In addition to acoustic sensors, they plan to deploy optic sensors to ensure the quality of the incoming tankers of piss.
This is interesting since its approach can be deployed in legacy equipment that doesn't have 'smart' sensors in it or deployed across shop floors that might have various pieces of equipment that don't 'talk' to each other or a centralized monitoring point. I work for such a place that has dozens of decades-old equipment that could benefit from such an approach. It is just too bad that Budweiser is spending all their money on cool manufacturing approaches and not on producing a drinkable beer!
Can you please stop submitting stories that link directly to paywalls?
KTHXBYE, fags.
So, when can we expect these in cars to detect failing hub bearing, waters bumps, idler pulley bearing, etc?
Life is not for the lazy.
I can recall reading several years ago (too lazy to find sources) where vibrometers or accelerometers (perhaps even audio) were used for the same thing. A working machine develops a healthy signature that you compare against. A mechanical EKG if you will.
A couple of microphones and a computer that compares soundbites is now "edge computing" and "machine learning."
Noise analysis of machines is neither new nor particularly exciting, and neither are wireless microphones.
Old news. This is a very common practice in factory automation.
Bud Miller and Coors low flavor beers have a place in the market too. They are OK served cold on a hot day. There are plenty of craft beers here near San Diego but many people can't afford or don't want to pay twelve or fifteen dollars for a six-pack.
In my day (submarines, 80's) it was a periodic check rather than real time (and definitely not wireless;-) ) It was considered valuable, because it really worked.
Motors, etc, all had little shiny disks glued on for the magnetic pickups; the sound guys got recordings and compared with previous ones (on paper.) They were working on a way to do it with reciprocating machines like compressors and maybe even diesels, which is probably possible now.
buzzword or not. These kinds of things are a sort of silent increase in efficiency that nobody's talking about. It's much quicker and cheaper to do the maintenance than it is to clean up a mess, but only if you know when to do the maintenance. Otherwise you're stuck spending a fortune on unnecessary maintenance.
This is literally an answer to the old Dilbert joke "I want advanced notice of any unplanned outages, and I want it yesterday". That sounded funny in the 90s, in 2019 somebody did it.
What worries me is when all these little efficiency boosts are applied on a global scale. Folks are gonna bring up the old "break windows to make jobs" hyperbole but that's not what this is. This is massively increasing the efficiency of maintenance done. Imagine if you never had to worry about a car part failing and leaving you stranded because you knew exactly when to do what and for how much. Imagine never paying a mechanic to fix something that didn't need fixing. That's what this is.
Hi! I make Firefox Plug-ins. Check 'em out @ https://addons.mozilla.org/en-US/firefox/addon/youtube-mp3-podcaster/
My very first job after graduating in 1981 was in acoustic emission, using ultrasound detectors (piezo-electric transducers) working in the 100 kHz to 2 MHz region to detect cracking in steel structures, at this point this technology was in regular NDT use, i.e. to verify lifting platforms, and people were starting to use them on rotating machinery.
It must be the use of machine learning to try to recognize the failure patterns which is the only thing new here.
Terje
"almost all programming can be viewed as an exercise in caching"
I mean this is a currently in development product by one of the largest and most common process control equipment manufacturers in the world. We've been trialing this with Emerson for 2 years already. It's quite a solid theory too: Build a signature of what your plant sound likes, detect changes to that signature and use wireless devices to triangulate.
I didn't realise that this has as much impact at a brewery but in hazardous industries the theory is solid: We've spent the past 20 years reducing the number of people in the plant so when things bang there are less injuries, but at the same time there's less ears in the plant able to detect a potential precursor to said bang.
Tassilo Festetics. It just rolls right off the tongue. Might be, I dunno, a brand name for stripper supplies or something.
Let's face it: American beer is the worst in the world. Like its chocolate - amazing that Americans are happy to eat Hershey's, which smells of puke.
Factories have been doing vibration analysis since the 90s.
I wrote an App for a company here in NZ which uses miniaturised Near InfraRed (NIR) spectroscopy to analyse and determine desired characteristics of organic and non-organic materials.
The biggest wins they have so far are in what is termed as "bio-prospecting" and one recent big win was in the Australian and NZ hop growers.
When you pick hops you can scan them using this device and print off a label with the date and the alpha value (bitterness) as well as other things of interest to brewers.
It gets even more interesting for buyers of old hop stocks because hops degrade over time. With this scanner you can tell what the current levels are.
It works using machine learning to build models from NIR "finger prints" of known samples.
There's a new data arms race that has already started and this one is mining the material world through sophisticated miniature sensors.
I predict that in 10 years NIR and other sensors will, together with sophisticated AI hardware and software, be a standard feature in every smart phone just as a camera is today.
Imagine being able to scan food or medicines with your phone to check for quality, allergies (nut content), authentication, age etc.
Low-cost sensors? Machine Learning? Maybe that's why Anheuser-Busch makes beers that taste like diet lemon-lime soda! Yuck! If I want decent beer, I hafta pay $2+ a bottle/can for a craft brew with a strange name. Anheuser-Busch should get outta the beer business. Fuck 'em.
You shouldn't be able to call anything "beer". Words have meaning. If there is more adjuncts than barely or contains polyvinylpolypyrrolidone, then it's piss.